Getty Images
Will data centers become obsolete?
Many organizations want to simplify or scale down their data centers -- but they won't disappear. Admins can examine as-a-service options and cloud to offload some applications.
Today's legacy data centers face unprecedented IT change and multi-pronged challenges. Initially, cloud services provided new levels of real-time scalability, compute power and storage. However, research indicates that C-suite leaders are beginning to retreat from these platforms due to data privacy and security concerns, as well as uncontrolled cloud spending.
Rising cloud costs have led many organizations to move cloud workloads back to on-premises environments. Enterprise data centers remain essential for ensuring the low latencies that edge deployments and AI workloads depend on. Increasingly, administrators and IT leaders are considering the advantages of hybrid deployments, including on-premises, public and private cloud, sustainable energy sources, smart automation and new hardware adoption.
However, research from Uptime Institute indicates that for the first time, less than half -- 48% -- of enterprise workloads are hosted in on-premises data centers. An increasing number of C-suite leaders are outsourcing these processing demands.
This article takes a close look at the relevance of legacy data centers and future changes, as well as strategies for adapting to new IT requirements, from IoT and edge to AI deployments.
Why do data centers still matter?
While the global demand for hyperscale data centers is expected to grow at a compound annual growth rate (CAGR) of 10.6% through 2030, reliance on today's enterprise data centers persists for several reasons. For example, processing data close to the source is important in healthcare, industrial IoT (IIoT) and financial markets. Real-time data analysis and nanosecond responses require edge or on-premises proximity to prevent latencies.
Increasingly, government and industry regulations mandate that data must be stored in specific regions, requiring information to remain local within data centers. As IT leaders continue to reduce their cloud spending, they've identified hybrid environments as economically and operationally vital. In many instances, dedicated facilities are important for ensuring strong data backup, failover and business continuity to maintain operational resilience.
What could cause data centers to become obsolete?
The energy requirements of AI workloads represent a serious challenge for legacy data centers. Data center electricity demand is projected to grow 16% in 2025 and double by 2030, according to Gartner.
AI requires rapid technical and operational changes that may render existing data center infrastructure less competitive or unsuitable for meeting future demands. The obsolescence of older hardware can also leave operators with underutilized assets. Interestingly, in terms of repatriation, a steady increase in operationalizing AI could lead to greater reliance on cloud providers for LLM training, AI deployments, and long-term management.
Common energy and resource bottlenecks in legacy data centers are detrimental to AI performance, and a greater reliance on GPU clusters could overwhelm existing power and cooling infrastructure. The result is C-suite and IT leaders turning to alternatives, such as major cloud hyperscalers, specialized AI platforms, and specific model and infrastructure providers.
The prospect of data center obsolescence also hinges on the steady emergence of new technologies that require significant Capex and rapid hardware refresh cycles. Given the two-to-three-year time frame for building new enterprise-level infrastructure, data center designs could be obsolete by the time construction begins.
Anticipated changes to the future of data centers
As data centers adapt to changes in resource requirements and a demand for edge capabilities, their infrastructure profiles will also evolve as AI workloads reshape data center designs and operations. For example, hybrid deployments that combine on-prem processing with private or public cloud will create a broad, distributed data center ecosystem consisting of hyperscale providers, colocation facilities and data-driven edge deployments.
New protocols and hardware will be necessary to drive increased energy efficiency, including renewable energy sources, liquid cooling and greater waste reduction. A report from Market.us points to significant global expansion in the liquid cooling sector, and in the U.S., the market is expected to grow at a CAGR of 17.1% through 2033.
Other changes include the drive toward intelligent automation to accelerate IT ops, increase high-bandwidth networking, and advance self-healing to repair equipment and network failures. Further, key aspects of workload deployments, IT management and security will all be automated in data centers of the future. According to McKinsey research, demand for AI-ready data center capacity is expected to grow annually at a CAGR of 33% through 2030, when AI workloads are expected to comprise 70% of total data center demand.
Challenges of maintaining on-premises data centers
Ed Featherston, Enterprise Architect and Independent Consultant with Osprey Software, said he felt any notion of widespread flight from data centers was "kind of naive and not reflecting reality."
However, he agreed that the days of businesses building, owning and maintaining their own data centers are waning. Simply put, that's because the time and resources required to operate and maintain those data centers are not core business activities. The challenges of operating data centers, particularly in finding talent, only grow more daunting, according to Featherston.
AI adoptions boost data center efficiency
Scott Sinclair, practice director of cloud, infrastructure and DevOps at Omdia, a division of Informa TechTarget, confirms that AI integration within on-prem data centers offers a critical opportunity to extend automation and simplify IT processes. Increasingly, AI capabilities are being adapted for IT ops and observability solutions. As preexisting data centers are retrofitted to support AI deployments, they offer new opportunities for reduced complexity and lower costs.
"In fact, 89% of organizations expect to leverage their budget for AI initiatives that will help modernize their infrastructure to better support not only AI, but other business-critical workloads as well," says Sinclair.
It appears that the automation of on-prem data centers is inevitable. And for those businesses that expect to support their own private AI initiatives, manual IT ops are simply unsustainable. Moreover, IT decision-makers want the greatest flexibility when it comes to deploying new technology or applications. This versatility extends to AI-driven management for greater energy efficiency, hybrid cloud frameworks to harness the benefits of private deployments and absolute control over sensitive proprietary data.
"According to our research, 76% of IT decision makers agree that they view on-premises application deployments more favorably today than they did five years ago," states Sinclair. "When business success is often derived from the strength of your digital capabilities, it's vital to have the flexibility to choose the best option for your application," he adds.
Competing IT priorities drive push to automated systems
When IT staff are so busy, they want to find ways to offload work, which is helping the as-a-service approach earn more attention.
Addressing numerous organizational needs means the old consumption model is no longer sustainable. People saw that as their job -- for example, being an expert in forecasting infrastructure needs. But now, with more automation, IT departments want to focus on other areas.
Part of the challenge organizations have in modernizing their on-premises infrastructure or offloading some applications frequently revolves around a lack of true cost visibility into their current on-premises facilities. For his part, Featherston takes exception to the cloud being the only, or even the primary, option.
"There is a significant move and focus to colocation facilities and managed services," Featherston said. There, as in the cloud world, organizations are starting to realize economies of scale in real estate, power, cooling and staffing. CoreSite research has shown that 98% of organizations are embracing or adopting a hybrid model that blends on-premises, colocation and cloud environments.
"When [colocation] and managed services prices are looked at, the pricing at first might be intimidating," Featherston said. "But those that do [look at] it find, ultimately, it is a much easier paradigm to manage."
Plan early and often to adapt data centers to new demands, including staff
Tracy Woo, senior analyst at Forrester Research, said the most important way to prepare for any changes ahead is to hire people with the right skills. For example, organizations moving to the cloud will need to find talent to support it. As IT roles compress, many will need to know how to code, manage infrastructure as code, and use automation and orchestration tools.
"It isn't about provisioning and providing services anymore," Woo said. "Much of that is done through self-provisioning, but it is more about integration and support activities." Moreover, security remains a concern for on-premises systems, and GPTZero research found that 54% of U.S. cybersecurity professionals use AI for network traffic monitoring.
Likewise, there used to be more functional silos, such as testing. Now, IT is more about platform teams and integrated teams. It isn't like the old "develop it and throw it over the wall" model, Woo said.
Because of that, traditional infrastructure management services teams need to know how to do continuous delivery and monitoring. Using observability, they need to figure out how to have visibility across the whole environment and how to use multi-cloud management tools.
Greg Schulz, founder of IT analyst and consulting firm StorageIO, urges practitioners to think broadly about their existing data centers and their potential value. Some data centers are particularly well located for reliable, affordable power and bandwidth.
"You can scale down your IT operations, but perhaps, you can use the facility for other things within the business," Schulz said. "There may be a high value to that facility that can help you or the business meet new goals."
Editor's note: This article was updated in February 2026 to reflect new data center infrastructure statistics and analysis aligned with popular trends, such as AI workloads and support, as well as edge deployments.
Kerry Doyle writes about technology for a variety of publications and platforms. His current focus is on issues relevant to IT and enterprise leaders across a range of topics, from nanotech and cloud to distributed services and AI.
Alan R. Earls is a Boston-based freelance writer focused on business and technology. He has done freelance work for publications ranging from CIO, Datamation and Computerworld to The Boston Globe, The Chicago Tribune, Modern Machining and Ward's Automotive.
